Workshop on information management for situational analytics

Situational Analytics It has long been argued that the fundamental purpose of is to enable managerial decision making. Prior to making decisions, managers often employ some form of analytics. Analytics could range from simple trend analysis through cause-effect modelling and sophisticated data mining. In every case, the decision to be made is about optimizing some organizational objective. This objective is influenced by activities in the organization. Therefore, once the objective has been accepted as important, the actual decision is not so much about the objective as it is about the activities that drive accomplishment of the objective. Each analytic situation therefore, can be characterized as an ensemble of objectives and drivers: the more explicit the linkages between these objectives and drivers the more understandable the decision. Thus the situation---the specific pattern of objectives and drivers---should determine the data to be used and the analytics relevant to the decision. Steps towards Situational Analytics The term situational generally refers to creating flexible, agile, adaptive, self-service capabilities that reduce the amount of time needed to update the data available in systems. This concept is fundamentally about granting capability to the information consumer to capture new data points as needs emerge, not necessarily to pattern the data into a decision framework that enables more rapid decision making. This patterning of data to fit the decision task is important in reducing the decision maker's cognitive load thus improving both the speed and quality of decisions. At the moment, when a decision maker interacts with BI-delivered data, he or she typically has to explore the data looking for trends and relationships between data points then consider the type of analytic technique that best fits with the data. Only then can the decision maker actually conduct the analytic process to make the decision. Information architectures that deliver the data in a way that fits the user's decision framework create a more explicit linking of data to the analytic situation. Therefore, while situational is about reducing the time to adaptation of the data, Situational Analytics is about reducing decision latency. An Architecture for Situational Analytics The architecture for Situational Analytics is based on the current 3-tiered architecture which includes a source system, a means of extracting data from the source and integrating it into a BI (which could include operational data stores, data marts, data warehouses, or tool-specific content storage areas) and a presentation layer where the data is delivered in different formats (multi-dimensional cubes, dashboards, or reports) to decision makers. Since the notion of situational analytics in goal-directed systems includes an ensemble of objectives and drivers, the architecture for situational analytics adds a modeling layer that allows for representation of the relationships between the objectives and drivers. In effect, the modeling layer replicates the framework by which the user makes decisions about how to improve performance relative to the pre-defined objectives. This layer also allows for identification of Key Performance Indicators (KPIs) to be used in the analysis. In addition, a link between the decision framework as represented in the modeling layer and the repository is required. That is, while the data mart or data warehouse is often designed using various data modeling techniques, the decision framework can be used as a means of selecting specific data points from those defined in the data model as key factors required for the situation. In this way, decision makers can reduce the amount of time they spend searching through data that might not be relevant to the decision, and focus on the relationships between key drivers and their attendant objectives. It is also a fact that most KPIs are influenced by business processes at different layers in the organization. Often the real reasons behind performance are hidden in business processes. Therefore, information about these processes are needed calling for a KPI-based but process-oriented view of the decision environment. That is, situational analytics requires access to aggregated KPIs but also a link from these KPIs to the process level data allowing decision makers to fully understand the impact of drivers on business objectives. Notwithstanding the importance of fully understanding the relationships between drivers and objectives, the growth of social media analytics begs for an integration of external data into the overall architecture of Situational Analytics. Typically, external sources can include a significant amount of unstructured data not easily captured in traditional tools. Business decisions however, often require a balancing of internal data (i.e., from Key Performance Indicators) and external data (i.e., customer comments or ratings on social media sites). Therefore, while decision frameworks that enable focus on internal KPIs are important elements of Situational Analytics, the ability to integrate external data is also a key factor. The critical issue here is to enable decision makers to understand the relative influence of different external data from competitors or from customers on their business objectives. Finally, modern organizations tent operate in a network of partners that might form and dissolve quickly once specific projects have been completed. A critical question for Situational Analytics is how do partners within a network share performance information in order to enable collaborative activity? Clearly, each partner will want to keep some information internal to that organization while sharing some data with others in the network. In these situations, an SOA approach can enable the sharing of decision frameworks and KPIs among partners thus facilitating the formation and management of productive networks and/or virtual organizations. In summary then, the architecture for Situational Analytics calls for tools that enable the modeling of decisions, integration of external data, and a view of the processes that contribute to KPIs thus providing a full snapshot of the decision environment whether within an organization or across partners within a network. Human Factors Notwithstanding the challenges of architecting a Situational Analytics system, a critical issue in any technology adoption curve within organizations is the capability of humans to fully make use of such systems itself. The logic of the Situational Analytics architecture is that the presentation of data fits more closely to the decision task at hand (mediated by the decision framework). Therefore, ease of use should be enhanced leading to more frequent exploitation of the capabilities of such a system. Over time, frequency of use should lead to the development of skills related to decision making with data. Core questions • What is the underlying logic of situational analytics? Is it about faster data updating or is it about reducing decision latency? Can we, in fact, improve decision quality and how would this be measured? • How does situational analytics integrate internal and external data? What are the underlying architectures needed for such integration? • What are the skills required of decision makers to make full use of such a system?

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.